The Quad ESL63 and its successors are classics of audio. The design, begun in 1963 and presented to the public as a finished product in 1981, was a sensation from the start and has been a contender for “the world’s best speaker” ever since—or, more precisely, the world’s best speaker in certain ways. The current 2812 model is the best version of the design to date in physical integrity. Its bass performance also seems superior to earlier versions. Sometimes, when family firms are sold and moved, the products go downhill in fit and finish. But this is not the case here. The new model is a work of the highest level of craftsmanship. One is almost tempted to call it a work of art. And it sounds essentially as one expects a Quad to sound. Going on memory alone, it seems to me to have preserved the essential Quad balance, though for some reason, it seems slightly brighter than the 2805, though I did not have a chance to compare the 2812 and 2805 side by side.
Before I go further, I want to point out that Chris Thomas did an online TAS preview (theabsolutesound.com/articles/quad-esl-2812-electrostatic-loudspeaker) which amounts to a thorough review in itself. I strongly recommend reading this as a supplement to what I have to say here, with a different emphasis.
When the ESL63 first came out long ago, HiFi Choice commented: “The sound was something of an acquired taste, [but] if its particular qualities appealed, they could assume such overriding importance that no other model [of speaker] would suffice.” I think it is fair to say this remains the case. There are many serious people in audio, our own Paul Seydor among them, for whom some version of the Quad ESL63 design remains the one and only. They may add subwoofers on occasion, some of them use DSP to fill in the lower mid and upper bass, and so on. But when the chips are down, the times may be a changin’ but their speaker is not.
Since Quad has had very limited distribution in the U.S. for a while, there may be a good many people who are not familiar with its distinctive and in many respects distinctively marvelous sound. For those of you in this situation, I am going to describe the sound ab initio. For people familiar with the design’s sound from earlier iterations, all you really need to know is that the latest version is very similar to former ones but in detail it is the best yet, I think—in particular, it is physically and sonically more solid.
How the Quads Work
The speaker looks like a single-diaphragm electrostatic, without a box, operated as a dipole. But it is not, in fact, a single unit in its operation. A small, inner, disc-shaped part is operated first. Then a ring around that central disc is fed the signal a bit later, the next ring out later still, and so on. (Physically, four panels fit together, each playing various parts in this concentric arrangement.) The idea that such an arrangement could be made to emit a spherical wavefront (to a very good approximation) was developed by Christiaan Huygens in 1678 (that is not a typo—late seventeenth century). It must have been embarrassing to Quad founder Peter Walker when, upon the first appearance of the ESL63, some audio reviewers described his genius in thinking of this wavefront-synthesis idea, which of course he well knew was old and standard physics. But regardless of previous theory, a great deal of ingenuity was required to make the thing a practical reality and the design is one of the signal achievements in audio history.
Nowadays one would run computer simulations. But back when, Walker described himself as working it out on the back of an envelope. The whole matter was not made any easier by the fact that in the analog world, pure time delays do not arise naturally. One has to make rather complicated cascaded passive networks for this. But it was all figured out, and the result was and is an approximation of a “virtual point source” about 30 centimeters behind the speaker (from either side!). The speaker does not, in fact, really sound like a point source when heard at close range. It sounds like a distributed radiator to some extent. But it does have a very consistent radiation pattern, effectively a true dipole up to a fairly high frequency range, above which the pattern is deliberately narrowed. A smoothly varying radiation pattern is a key to low coloration (something which Walker clearly understood long before Floyd Toole publicized this). And low coloration is exactly what Walker got.
A dipole panel of the size here (effectively two feet square) starts to roll off considerably above the deep bass. This roll-off is offset in part by a resonance around 80Hz. It also helps that the speaker is on the floor so part of the backwave is blocked from coming around to the front (in reverse polarity). But deep bass is not the specialty of the Quads even so. Below 50Hz or so, they are on their way out, though the bass down to where they roll off is quite convincing and defined and clean.
As electrostatics tend to be, the Quads are very low in distortion overall. Quad used to advertise proudly that THD (total harmonic distortion) was under 0.1% from 100Hz on up. This amounts to distortion at least 60dB down from signal. The present manufacturer’s “specs” on this are a bit higher, but are quoted at a 100dB level—very high—and the figures are still low. Even in the bass, in the 50 to 100Hz octave, the distortion is under 1%, a figure few box speakers can equal.
For comparison, most dynamic-driver speakers are happy to get under 45dB down across the board, though some few get down consistently to close to Quad levels. The Infinity Primus 363s are 50dB or more down (above the deep bass), for example. And the Cerwin Vega CLS 215s have bass distortion under 1%, even at high levels.